Nvidia Design Partner Arm Says It's Going to Sell Its Own Chips. Its Stock is Soaring.

Arm unveiled its first in-house AI chip, the AGI CPU, and guided to $25B in revenue by 2031 with $15B from its own chip sales and $9 adjusted EPS in 2031. Shares surged about 20% to roughly $161 following the announcement; Meta is the AGI CPU's first customer. Raymond James upgraded to outperform with a $166 price target while Bank of America raised its target to $155 but remained neutral, and Visible Alpha shows current fiscal-year revenue near $5B and adjusted EPS of $1.75.

Analysis

Arm moving from pure-IP to productized silicon creates an explicit vertical tension with its largest customers that will drive two opposite responses: hyperscalers accelerate bespoke core/licence work to avoid dependence, and a tranche of mid-tier cloud/service providers will outsource to Arm to close capability gaps faster than building internal design teams. That bifurcation amplifies demand for foundry capacity at leading nodes for bespoke designs while simultaneously enlarging the addressable market for turnkey Arm-branded systems — a structural win for foundries, advanced packaging, and HBM channel dynamics even if unit ASPs compress initially.

Execution hinges on three non-linear levers: performance-per-watt parity with incumbent custom cores, credible software/toolchain parity (compilers, libraries, orchestration for AI stacks), and predictable wafer supply agreements. Any one failure creates an outsized reversal: poor tapeout metrics or foundry delays will cause hyperscalers to accelerate migration to alternate ISAs or internal cores, shrinking Arm’s licensing pool and compressing forward valuation multiples within quarters rather than years.

For portfolio construction, treat this as a multi-year optionality sale/roll situation rather than a straight growth multiple expansion trade. Short-term sentiment can remain extrapolative; medium-term value accrues only if Arm secures multi-hyperscaler adoption, foundry capacity, and software ecosystem lock. Therefore phase sizing, volatility monetization, and explicit hedges tied to hyperscaler behavior are the right playbook — the asymmetric payoff is real but concentrated in a handful of binary delivery and customer-adoption events over the next 12–36 months.

Key Decisions for Investors

• ARM — Buy an 18–30 month call spread (LEAP debit call spread) sized 1–3% of NAV to capture multi-year adoption optionality; finance by selling a higher-strike LEAP to reduce cost. Target 2x–3x upside if Arm secures two-plus independent hyperscaler design wins within 24 months; max loss is premium paid (~1–3% NAV).
• ARM — Sell 30–90 day call spreads (delta ~20/35) sized 0.5–1% NAV to harvest elevated implied volatility ahead of technical milestones; close on public tapeout/benchmark windows or if stock gaps >10% on event days. Use proceeds to fund LEAP purchases.
• META — Buy 9–15 month call (or add 1–2% equity exposure) as a low-cost way to play early adopter upside from integrating Arm silicon into its stack. Risk: pilot underperformance; reward: 2–3x if Meta scales internal deployments. Trim half on missed integration KPI at next public milestone.
• NVDA — Buy 6–9 month downside protection (puts) sized to cap tech-beta exposure from ARM directional trades (0.5–1% NAV). Cost ~25–75 bps of NAV but protects against a sector-wide re-rating should Arm trigger licensee defections or broader demand concerns.